<p>This paper addresses key challenges of severely exacerbated stiffness anisotropy and severe cable tension imbalance in Cable-Driven Parallel Robot (CDPR) under large external moments, where conventional methods—designed primarily for free of payload—fail to prevent the persistent tension split between cable groups, leading to compromised stiffness uniformity. To tackle this problem, we propose a real-time tension distribution algorithm for a 6-degree-of-freedom (DOF) CDPR actuated by 8 cables. First, we establish the system’s kinematic and stiffness models to analyze how tension distribution influences stiffness anisotropy. To mitigate the impact of large external moments, we formulate the Balanced Stiffness Optimization (BSO). This approach aims to prevent cable overload while enhancing overall stiffness homogeneity. Furthermore, to ensure numerical stability and real-time performance, we propose the Condition-Adaptive Real-time Solver (CARS). This method maintains millisecond-level solving speeds and effectively handles failures caused by an ill-conditioned structure matrix. Simulation and experimental results demonstrate that our method achieves continuous, smooth tension distribution, significantly reduces the disparity between high-tension and low-tension cable groups, and improves overall system stiffness. The proposed algorithm outperforms traditional methods in both real-time capability and robustness, providing a reliable technical foundation for CDPR operations under large external moments.</p>